mlsanityData to 14 Oct 2026

Revisiting Essential and Nonessential Settings of Evidential Deep Learning

arXiv 2410.00393 ↗First posted 1 Oct 2024cs.LG cs.AIcited by 15

Not on the map. Why?

Abstract (arXiv)

Evidential Deep Learning (EDL) is an emerging method for uncertainty estimation that provides reliable predictive uncertainty in a single forward pass, attracting significant attention. Grounded in subjective logic, EDL derives Dirichlet concentration parameters from neural networks to construct a Dirichlet probability density function (PDF), modeling the distribution of class probabilities. Despite its success, EDL incorporates several nonessential settings: In model construction, (1) a commonly ignored prior weight parameter is fixed to the number of classes, while its value actually impacts the balance between the proportion of evidence and its magnitude in deriving predictive scores. In model optimization, (2) the empirical risk features a variance-minimizing optimization term that biases the PDF towards a Dirac delta function, potentially exacerbating overconfidence. (3) Additionally, the structural risk typically includes a KL-divergence-minimizing regularization, whose optimization direction extends beyond the intended purpose and contradicts common sense, diminishing the information carried by the evidence magnitude. Therefore, we propose Re-EDL, a simplified yet more effective variant of EDL, by relaxing the nonessential settings and retaining the essential one, namely, the adoption of projected probability from subjective logic. Specifically, Re-EDL treats the prior weight as an adjustable hyperparameter rather than a fixed scalar, and directly optimizes the expectation of the Dirichlet PDF provided by deprecating both the variance-minimizing optimization term and the divergence regularization term. Extensive experiments and state-of-the-art performance validate the effectiveness of our method. The source code is available at https://github.com/MengyuanChen21/Re-EDL.

Key facts. Cited by 15 papers in the map (data through 14 Oct 2026). Builds on 61 of its 115 references that are in the map. First posted on arXiv 1 Oct 2024.

Builds on

61 of 115 references.

References · 61

  1. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,448
  2. 2017Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithmscited by 4,349
  3. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,926
  4. 2016Matching Networks for One Shot Learningcited by 2,255
  5. 2018Evidential Deep Learning to Quantify Classification Uncertaintycited by 732
  6. 2018Predictive Uncertainty Estimation via Prior Networkscited by 557
  7. 2015Towards Open Set Deep Networkscited by 538
  8. 2018Deep Learning for Classical Japanese Literaturecited by 422
  9. 2016Tutorial on Variational Autoencoderscited by 409
  10. 2020Simple and Principled Uncertainty Estimation with Deterministic Deep Learning via Distance Awarenesscited by 389
  11. 2020BatchEnsemble: An Alternative Approach to Efficient Ensemble and Lifelong Learningcited by 371
  12. 2019Deep Evidential Regressioncited by 355
38 more
  1. 2021What Are Bayesian Neural Network Posteriors Really Like?cited by 308
  2. 2020Efficient and Scalable Bayesian Neural Nets with Rank-1 Factorscited by 171
  3. 2022Trusted Multi-View Classification with Dynamic Evidential Fusioncited by 168
  4. 2021Free Lunch for Few-shot Learning: Distribution Calibrationcited by 165
  5. 2020Posterior Network: Uncertainty Estimation without OOD Samples via Density-Based Pseudo-Countscited by 164
  6. 2020Learning Open Set Network with Discriminative Reciprocal Pointscited by 115
  7. 2019Gaussian YOLOv3: An Accurate and Fast Object Detector Using Localization Uncertainty for Autonomous Drivingcited by 105
  8. 2019Reverse KL-Divergence Training of Prior Networks: Improved Uncertainty and Adversarial Robustnesscited by 104
  9. 2021Evidential Deep Learning for Open Set Action Recognitioncited by 96
  10. 2022Pitfalls of Epistemic Uncertainty Quantification through Loss Minimisationcited by 66
  11. 2023Uncertainty Estimation by Fisher Information-based Evidential Deep Learningcited by 58
  12. 2020Uncertainty-Aware Deep Classifiers using Generative Modelscited by 41
  13. 2022The Unreasonable Effectiveness of Deep Evidential Regressioncited by 40
  14. 2021Trustworthy Multimodal Regression with Mixture of Normal-inverse Gamma Distributionscited by 39
  15. 2023On Second-Order Scoring Rules for Epistemic Uncertainty Quantificationcited by 38
  16. 2019Multi-Moments in Time: Learning and Interpreting Models for Multi-Action Video Understandingcited by 37
  17. 2023Second-Order Uncertainty Quantification: A Distance-Based Approachcited by 35
  18. 2021AutoDEUQ: Automated Deep Ensemble with Uncertainty Quantificationcited by 31
  19. 2022Post-hoc Uncertainty Learning using a Dirichlet Meta-Modelcited by 31
  20. 2023Single-model uncertainty quantification in neural network potentials does not consistently outperform model ensemblescited by 30
  21. 2023Learn to Accumulate Evidence from All Training Samples: Theory and Practicecited by 29
  22. 2022Towards Open Set Video Anomaly Detectioncited by 26
  23. 2023Open Set Action Recognition via Multi-Label Evidential Learningcited by 26
  24. 2023ELFNet: Evidential Local-global Fusion for Stereo Matchingcited by 25
  25. 2021Multivariate Deep Evidential Regressioncited by 23
  26. 2022OpenTAL: Towards Open Set Temporal Action Localizationcited by 23
  27. 2023Prototype-based Aleatoric Uncertainty Quantification for Cross-modal Retrievalcited by 23
  28. 2023Revisiting the Domain Shift and Sample Uncertainty in Multi-source Active Domain Transfercited by 20
  29. 2023Second-Order Uncertainty Quantification: Variance-Based Measurescited by 19
  30. 2024Are Uncertainty Quantification Capabilities of Evidential Deep Learning a Mirage?cited by 18
  31. 2020Multidimensional Uncertainty-Aware Evidential Neural Networkscited by 16
  32. 2023Evidential Uncertainty Quantification: A Variance-Based Perspectivecited by 14
  33. 2019Quantifying Classification Uncertainty using Regularized Evidential Neural Networkscited by 13
  34. 2021Natural Posterior Network: Deep Bayesian Uncertainty for Exponential Family Distributionscited by 13
  35. 2021Improving evidential deep learning via multi-task learningcited by 13
  36. 2023Exploring and Exploiting Uncertainty for Incomplete Multi-View Classificationcited by 13
  37. 2024Uncertainty Regularized Evidential Regressioncited by 13
  38. 2023Continual Evidential Deep Learning for Out-of-Distribution Detectioncited by 11

50 most cited of 61 All 61 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 1; 2025: 2; 2026: 12.

15 papers.

Sort:

By area:

  1. 2024Conjugated Semantic Pool Improves OOD Detection with Pre-trained Vision-Language Modelscited by 13
  2. 2026Revisiting Uncertainty: On Evidential Learning for Partially Relevant Video Retrievalcited by 2
  3. 2025An Uncertainty-Aware Dynamic Decision Framework for Progressive Multi-Omics Integration in Classification Taskscited by 1
  4. 2025Bi-level Meta-Policy Control for Dynamic Uncertainty Calibration in Evidential Deep Learningcited by 1
  5. 2026GEM-FI: Gated Evidential Mixtures with Fisher Modulationcited by 1
  6. 2026Density-Informed Pseudo-Counts for Calibrated Evidential Deep Learningcited by 0
  7. 2026Ensemble-Based Dirichlet Modeling for Predictive Uncertainty and Selective Classificationcited by 0
  8. 2026Are Independently Estimated View Uncertainties Comparable? Unified Routing for Trusted Multi-View Classificationcited by 0
  9. 2026Rethinking Vacuity for OOD Detection in Evidential Deep Learningcited by 0
  10. 2026Plug-in Losses for Evidential Deep Learning: A Simplified Framework for Uncertainty Estimation that Includes the Softmax Classifiercited by 0
  11. 2026Courtroom Analogy: New Perspective on Uncertainty-Aware Classificationcited by 0
  12. 2026Variational Inference for Evidential Deep Learningcited by 0
3 more
  1. 2026Spatio-Temporal Fusion Model for Standard View Classification of Echocardiographic Videoscited by 0
  2. 2026UASPL: Uncertainty-Aware Self-Paced Learning with Evidential Neural Networkscited by 0
  3. 2026Uncertainty-driven training for three-dimensional calibrated lung nodule classificationcited by 0